Opens in a new tab

Home / Resources / AI in Research

AI-Assisted Analysis in Doctoral Research:
What Your Committee May Ask

AI use in doctoral research can be productive, especially for generating syntax drafts, checking documentation, clarifying statistical concepts and planning reproducibility checks. The key is to follow the university's actual rules, protect research data, and be able to explain and defend every methodological choice in the doctoral project (Kasneci et al., 2023; van Dis et al., 2023).

AI in Research
WHO ASKSCommittee, university
WHEREMethods, disclosure
KEEPProvenance record
NEVERPaste unapproved data
LAST CHECKEDNovember 2026

A doctoral examination tests the researcher’s reasoning, not how little technology they used. AI assistance should be documented in a way that preserves that reasoning and makes the final analysis auditable.

AI use in doctoral research: where is it useful?

POLICY TABLE · Last checked November 2026 · rechecked every six months

Research activity Useful assistance Doctoral researcher still needs to establish
Research planning Compare candidate designs; draft a decision checklist Research question, design justification and assumptions
Data preparation Propose recoding rules and syntax Whether variables mean what the code assumes
Data analysis Draft statistical scripts and diagnostic checks Model fit, prespecified analyses and valid inference
Results writing Explain statistical terminology and suggest structure Exact results, interpretation and consistency with tables
Revision Organize committee comments into an action log Appropriate and accurate responses to each critique

These are example task divisions, not blanket authorizations to upload doctoral research data to a commercial tool.

Five questions a doctoral committee may reasonably ask

1. Why was this method selected? The explanation should start with the research design, outcome scale, dependence structure and target comparison. An AI suggestion is not a substitute for that methodological argument.

2. How were the data prepared? Present transparent variable definitions, reverse coding, exclusions, missingness rules and any departures from the original protocol.

3. Can the analysis be rerun? Preserve executable syntax, software versions, codebooks and a table-to-output mapping. Sandve et al. (2013) explain why reproducible computational work depends on these artifacts rather than a narrative description alone.

4. What did AI actually do? Identify whether it suggested code, drafted text, summarized nonsensitive materials or directly processed data. Different actions may trigger different university policies.

5. Who checked the results? The candidate should be able to describe method checks, diagnostics and interpretation. A statistician’s review can improve confidence, but responsibility for the scholarly claims remains with the researcher.

The five questions above are an editorial preparation guide, not a universal university examination protocol.

Is AI assistance the same as academic misconduct?

No universal answer applies. Universities and programs can distinguish permitted brainstorming, coding support and editing from prohibited outsourcing or unacknowledged content generation. Policies and assessment requirements vary. Consult the graduate school’s current guidance, the IRB protocol and the doctoral committee’s instructions for the specific use case.

Research on educational uses of language models discusses both opportunities for individualized support and concerns about assessment, knowledge quality and responsible use (Kasneci et al., 2023). Neither the benefits nor the concerns support making up a blanket rule for every university.

What should be documented from an AI-assisted analysis?

Keep an analysis provenance record: research question; approved protocol and analysis plan; AI tool and relevant version if required; task requested; code accepted or rejected; edits; software and package versions; data checks; and the final executed results. Retain only information that institutional rules allow, avoiding duplication of sensitive participant records in AI logs.

If a model suggests an alternative test, record whether that test was part of the original plan, why it was considered, and whether the change was exploratory. This prevents an attractive new result from quietly replacing the analysis planned before results were seen.

Example doctoral research disclosure wording

Illustrative wording; adapt to institutional rules and actual work:

EXAMPLE WORDING (ADAPT TO THE JOURNAL)

A generative AI tool was used to assist with drafting and debugging statistical syntax. The final analysis was performed in [software/version] using the study dataset in the authorized research environment. The researcher verified variable definitions, analytical decisions and outputs; [describe any independent methodological review]. No participant-level data were entered into an external AI service [include only if true].

A disclosure should never describe checks that did not happen. Research authorship and responsibility principles are discussed in the ICMJE recommendations for medical research, although doctoral research rules come from the institution (ICMJE, 2026).

Frequently asked questions

It can explain what a statistic generally means. Verify the result against the original output and avoid inferring causation from an observational association without a defensible design.

Not every university requires one. Independent review may be useful for complex designs, repeated measures, mediation, prediction models or consequential reviewer comments.

Not until the data steward and applicable university/IRB permissions permit that specific processing arrangement. Redacting names alone may be insufficient.

Independent Statistical ReviewNeed a statistician to confirm the analysis before your committee sees it? An in-house PhD-level statistician reviews the analysis, while the research question, interpretation and conclusions remain yours.

Request an Assessment

Independent Statistical Review

Independent Statistical Review at OnlineSPSS is carried out in-house by PhD-level statisticians, and it checks method, assumptions, reproducibility and reporting. The research question, interpretation and conclusions remain yours. Independent Statistical Review covers analyses prepared with AI assistance.

References

  • International Committee of Medical Journal Editors. (2026). Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals: Use of AI by authors. https://www.icmje.org/recommendations/browse/artificial-intelligence/ai-use-by-authors.html
  • Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., … Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, 103, 102274. https://doi.org/10.1016/j.lindif.2023.102274
  • Sandve, G. K., Nekrutenko, A., Taylor, J., & Hovig, E. (2013). Ten simple rules for reproducible computational research. PLOS Computational Biology, 9(10), e1003285. https://doi.org/10.1371/journal.pcbi.1003285
  • van Dis, E. A. M., Bollen, J., Zuidema, W., van Rooij, R., & Bockting, C. L. (2023). ChatGPT: Five priorities for research. Nature, 614, 224–226. https://doi.org/10.1038/d41586-023-00288-7

Last reviewed November 2026 by the OnlineSPSS Statistical Team. Worked examples use hypothetical values and are not client data.

NEXT STEP

Need this done for
your study?

Share your research in about a minute and receive a complimentary assessment the same day: scope, timeline and one fixed fee.

No Commitment Required

Strictly Confidential

Fixed Project Fee

In-House PhD-Level Statisticians

Same-Day Response

Serving Researchers Since 2014